PHONOLOGICAL REPRESENTATION

A mental representation of the individual phonemes of a language or of the phonological forms of its words, which enables the language user to recognise and to produce them.

There is a lay assumption that the sounds of a language are recognised by a simple process of matching them on a one-to-one basis with some kind of template in the mind. This cannot be the case, due to the high variability of phonemes in the speech signal. Co articulation causes the form of any phoneme to vary considerably according to the phonemes which precede and follow it. It has also proved impossible to slice the signal into neat phonemic segments since acoustic features representing one phoneme blend into those representing the next. There is enormous variation between speakers in terms of pitch of voice, speech rate and accent. There is also great variation within the speech of a single speaker reflecting their emotional state and the genre of speech event. To this, one can add features such as assimilation, where the speaker modifies phonemes in the interests of ease of articulation.

Broadly, there are three ways of approaching the issue of phonological representation. One can assume an intermediate level of analysis (solution a below), one can adapt one’s view of how the knowledge is stored (solutions a, b, d and e below) or one can adapt one’s view of how the user processes the information in the signal (solution c):

Unit of representation. Representation may be at a higher or lower level than the phoneme: for example, at the level of the syllable or the acoustic feature. The analysis of the signal would then involve an intermediate process of dividing it into these units.

The case for a syllabic representation is that phoneme values within the syllable remain relatively constant, thus countering the co-articulation problem. It has been calculated that for French a finite set of 6000 syllables could represent the entire repertoire of the language. Demi-syllables would be even more efficient, requiring perhaps 2000 forms.

 An alternative suggestion is that words in the lexicon might be represented in terms of smaller units than the phoneme, i.e. as combinations of acoustic features (+ voiced, + nasal etc.). This has much in common with the feature theories which have been put forward to explain other aspects of pattern recognition.

Prototype solutions. The mind stores an idealised version of each phoneme of the language (or of the phonological form of each lexical item). Evidence in the input can then be matched to the representation on a relative ‘goodness of fit’ basis rather than a categorical one. The prototype might consist of a core value, with a range of tolerances around it. These tolerances would incorporate systematic deviations from the prototype which exposure to variants (e.g. local accents) has taught the listener to recognise.

Normalisation. The listener edits out or adapts any information in the signal which does not serve to identify phonemes. Allowance is made for systematic variation caused by co-articulation and assimilation. Normalisation has especially been invoked to explain how the listener succeeds in recognising the same phonemes spoken by a variety of speakers. In its strong form, the theory postulates that phonological information is strictly separated from indexical information (relating to the characteristics of the speaker’s voice). This view has been challenged by recent evidence suggesting that the two are more closely interlinked than was supposed.

 Under-specification. An under-specified phonological representation contains only information that is critical to the identification of a lexical item. This means that absent features (e.g. ‘not nasal’) are not recorded; it also means that features which can be predicted by rule (e.g. the nasalisation of a vowel before [n]) are omitted. An under-specified representation reduces the chances that a match will fail because of extraneous details in the speech signal.

 Exemplar theories. Recent evidence has suggested that our capacity for storing information relating to individual events may be far greater than supposed. Applied to phonology, this means that the representation of a particular phoneme or lexical item may be composed of multiple traces of earlier encounters with that phoneme or item. When we hear a regional accent, we tap in to stored exemplars of items spoken in that accent; when we hear a voice operating at a certain pitch or speech rate, we can relate it to earlier stored experiences of similar voices. An exemplar approach is not necessarily incompatible with theory b: it may be that the language user also builds up a set of core values based upon the multiple traces that he/she has stored.

See also: Exemplar models, Normalisation, Pattern recognition, Prototype Theory, Under specification, Unit of perception

 Further reading: Bybee (2001: Chaps 1–2); Fitzpatrick and Wheeldon (2000)

اخر الاخبار

اخبار العتبة العباسية المقدسة

شعبة الخطابة الحسينية النسوية تستأنف برامج رابطة خطيبات المنبر الحسيني

شعبة فاطمة بنت أسد تحتفي بتخرج الدفعة الثانية من طالبات المدرسة القرآنية

مع انطلاق العام الدراسي.. العتبة العباسية المقدسة تطّلع على سير العملية التعليمية في مجموعة العميد التربوية

بأربع أوراق بحثية.. قسم الشؤون الفكرية يشارك في المؤتمر العلمي الدولي الرابع عشر بالعتبة الكاظمية

المزيد

الآخبار الصحية

كيف يمكن لحفنة من الجوز أن تؤثر على نومك وتركيزك؟

دراسة: كوب ميلك شيك واحد قد يكون "قنبلة دماغية"

هل يعد "الباراسيتامول" خيارا آمنا للنساء الحوامل؟

دراسة تكشف الرابط بين العلاقات العاطفية والرضا عن الحياة

المزيد

الآخبار التكنلوجية

كارثة صامتة.. "الهواء الملوث" يسرق نور عيون الأطفال

دراسة: رصد أول دليل على وجود "إشعاع هوكينغ"

رسالة بلغة لا تعرفها؟.. "واتساب" يترجمها لك فورا

"إنستغرام".. ميزة جديدة لكشف "الكذب في العمر"

المزيد

مواضيع ذات صلة

PHONOTACTIC RULES

PHONOLOGICALWORKING MEMORY

PHONOLOGICAL DEVELOPMENT: PRODUCTION

PHONOLOGICAL DEVELOPMENT: PERCEPTION

PHONOLOGICAL BIAS TECHNIQUE

PHONOLOGICAL AWARENESS

PERCEPTUAL MAGNET EFFECT

PERCEPTION

PAUSING

PATTERN RECOGNITION (also PATTERN MATCHING)

PARAPHASIA

PARALLEL PROCESSING